Best Practices and Strategies to Create Effective Visualizations

Aaron R. Williams

@awunderground

Best Practices and Strategies to Create Effective Visualizations

Why I Use R

Aaron Williams

@awunderground

Outline

1. WHY: Principles for selecting a data viz tool
2. WHAT: R and ggplot2
3. HOW: The theory behind R and ggplot2
4. Examples
5. Grade my decision

1. Looks good!

Source: fivethirtyeight

2. Flexible and expressive

3. Reproducible

4. Scalable

AKA “Please just don’t make me point-and-click 100 charts”

5. In my analysis workflow

Source: Hadley Wickham

Source: ggplot2

fivethirtyeight

Financial Times

In the first edition of @FT's World Cup Extra series of stats-y briefings, @muradahmed and I ask: as curtain-raisers go, just how underwhelming is Russia vs Saudi Arabia?https://t.co/0aQXcaD9YV pic.twitter.com/GOXVyVlRW7

— John Burn-Murdoch (@jburnmurdoch) June 13, 2018

BBC

BBC R Cookbook

Leland Wilkinson

Grammar of Graphics

Hadley Wickham

A Layered Grammar of Graphics

Source: Hadley Wickham

1. Data

ggplot(data = )

2. Aesthetic mappings

aes(x = , y = , color = )

X or Y

Color or Fill

Aesthetic mappings

Size

Shape

Others: transparency, animations, faceting

3. Geometric objects

Plots are often called their geometric object(s).

Bar

geom_bar()

3. Geometric objects

Line

geom_line()

3. Geometric objects

Point

geom_point()

3. Geometric objects

?

3. Geometric objects

Path of Hurricane Katrina

4. Scales

Before

scale_x_continuous()

After

scale_x_log10()

4. Scales

Before

scale_size_continuous(breaks = c(25, 75, 125))

After

scale_size_continuous(range = c(0.5, 20), breaks = c(25, 75, 125))

5. Coordinate systems

Cartesian

5. Coordinate systems

Polar

coord_polar()

6. Facets (optional)

Facet wrap

facet_wrap(~category)

6. Facets (optional)

Facet grid

facet_grid(month~status)

7. Statistical transformations (optional)

Count

8. Themes

Default theme

8. Themes

fivethirtyeight theme

8. Themes

urbnthemes

Coding example

New? 9. Arrangements

New? 9. Arrangements

Source: Policyviz by Jon Schwabish

New? 9. Arrangements

Animation

By Sarah Strochak

10. Interactivity?

urbnmapr

Pension code

3.2, 4.2, 4.3, 4.7, 5.12

R Shiny

https://www.urban.org/policy-centers/cross-center-initiatives/program-retirement-policy/projects/dynasim-visualizing-older-americans-future-well-being/fiscal-health-social-security

Graphics guide

1. Looks good!

A

2. Flexible and expressive

B

Once you understand ggplot2, it is really useful for doing 90% of the results with 10% of the effort

(Jennfer Hill)

3. Reproducible

A+

4. Scalable

B+

5. In my analysis workflow

A+